This study presents a multi-objective optimization framework integrating genetic algorithms with deep learning for spacecraft project management, addressing critical challenges in schedule prediction and cost forecasting. The methodology employs binary tournament selection with Pareto dominance ranking, simulated binary crossover ( \(p_c = 0.9\) , \(\eta _c = 20\) ), and CNN-LSTM hybrid architecture for temporal dependency modeling. Comparative analysis across six prediction methods demonstrates superior performance: RMSE of 34.7 days, MAPE of 8.3%, and R \(^{2}\) of 0.847, achieving 79.3% accuracy within ±30-day windows and 92.6% within ±60-day windows. The framework attains 94.7% accuracy for baseline expenditure prediction and 73.4% accuracy for unplanned cost growth forecasting, substantially outperforming traditional parametric approaches (25–40% accuracy). These results validate the framework’s capability to support data-driven resource allocation, risk-informed budget reserve planning, and proactive stakeholder communication in complex aerospace development programs, enabling more resilient project execution strategies. The framework demonstrates 73.4% accuracy in cost growth prediction compared to traditional 25-40% accuracy, enabling evidence-based decision support for billion-dollar aerospace initiatives.